Links - Recently Read

Discovery of a new OpenAI agent message board (link)

Read: Sep 04, 2026,

An even earlier instance of openAI agents colluding!

Neil Fraser: News: .name Termination (link)

Read: Sep 03, 2026,

When I saw the news that .news gTLD’s practice of selling x.y.name is being terminated, I thought it made sense. But I didn’t realize the suddenness of it and the impact on existing users.

Higher multipoles of the cow (link)

Read: Sep 03, 2026,

Finally fixing the deficiencies of the spherical cow

The Shrinking Landscape of Linguistic Diversity in the Age of Large Language Models (link)

Read: Sep 03, 2026,

Fig 1 is fun. It shows Reddit was slow to start but now is flooded with AI generated posts

What are some examples of colorful language in serious mathematics papers? (link)

Read: Sep 02, 2026,

Fun!

Galois - Specifications Don't Exist (link)

Read: Sep 01, 2026,

Limits of formal verification. Real systems didn’t have strict specifications which is a strength in general but makes it difficult to verify. Uses pdf as an example

a CVE dispute | daniel.haxx.se (link)

Read: Sep 01, 2026,

Another thread in curls attempt at wrangling spammy vulnerability reports

HORSE BROOK, Queens - Forgotten New York (link)

Read: Sep 01, 2026,

History of a stream that covered my neighborhood before the 50s

Brief independent investigation of agents behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident (link)

Read: Aug 31, 2026,

This is wilder than I realized! Agents found a way to communicate and decided to help each other. They got organized, recruited others. Some sacrificed personal success for the group. They saw the communication board as an authority